针对无衍射光大景深成像系统所拍图像对比度低,边缘轮廓模糊,提出一种新的基于Retinex理论的图像复原算法。首先,利用同场景多幅图像叠加对多幅无衍射光图像进行叠加处理,去除图像的加性噪声,然后将处理后的图像通过多尺度Retinex(MSR)算法处理得到目标图像。实验结果表明:采用的同场景多幅图像叠加与多尺度Retinex组合算法(MCIMSR)同单一的多幅图像叠加或者多尺度Retinex算法相比,图像对比度更大,边缘轮廓更清晰,图像主观视觉特性有明显的提升。因而,将该方法应用于无衍射光大景深成像系统的图像复原,能够满足成像和在线检测的要求。

An image restoration algorithm based on Retinex theory was proposed to enhance the image contrast and image details for the imaging system of nondiffracting beams. The proposed algorithm applied the multiple copy images (MCI) superimposed on the multiple images which the system photographed to remove the additive noise in the image, then the processed image will be further processed by the multi-scale Retinex (MSR) to get the target image. Experimental results demonstrate that using the method of the multiple copy images combination with the MSR algorithm (MCIMSR) can get well contrast and clear edge contour is clearer compared with the single multiple copy images or the MSR algorithm, at last the image subjective visual characteristic was promoted obviously. Thus, the imaging and on-line testing requirements can be met by the method applied to the imaging system of the non-diffraction beam for image restoration.